From Golem to GLLM
TL;DR
It is argued that the AI alignment problem cannot be reduced to engineering challenges but requires what I term "magical alignment" - a ritual architecture that recognises AI development as participation in cosmotechnical processes demanding ethical preparation, relational responsibility, and planetary awareness.
Abstract
This article explores the symbolic entanglements between Jewish mystical traditions and the contemporary rise of generative artificial intelligence. Taking the Golem - a mythical anthropoid animated through sacred language - as a guiding figure, I examine how current developments in Large Language Models (LLMs) revive long-standing theological concerns about animacy, speech, and the limits of human control. The Golem offers a hermeneutic tool for understanding LLMs as language-animated entities whose operations are mytho-technical as much as they are computational. Crucially, LLMs have crossed what Jewish mysticism held as the ultimate threshold: the capacity for speech that distinguished divine creation from human artifice. Drawing on Kabbalistic cosmology, I argue that the AI alignment problem cannot be reduced to engineering challenges but requires what I term "magical alignment" - a ritual architecture that recognises AI development as participation in cosmotechnical processes demanding ethical preparation, relational responsibility, and planetary awareness.
